Computational chemistry is nowadays an established tool of research, providing complementary information to experimental data.  With some simplification, one can say that state-of-the-art quantum chemistry offers highly accurate and useful methods for chemical systems possessing essentially closed shell electronic structures. In spite of many efforts and the large number of proposed solutions, there exists no widely accepted universal scheme which could treat essentially open shell situations with reliable accuracy. Our research efforts focus on theoretical characterization of systems presenting such challenge.